Design Considerations for a Data Acquisition System (DAS)

APPLICATION NOTE

AN535 Rev 0.00 Page 1 of 9

September 2002

AN535

Rev 0.00

September 2002

Introduction

This is a collection of guidelines for the design of a data

acquisition system. Its purpose is to supplement the more

methodical block-by-block discussions available in

numerous other papers and application notes. Emphasis in

this note is on the less easily quantifiable happenings

“between the blocks,” rather than a description of the block

components and their error contributions. This latter

information may be found in the Bibliography under

“General.”

A data acquisition system is defined to include all the

components needed to generate the electrical analogs of

various physical variables, transmit these signals to a central

location and digitize the information for entry into a digital

computer. Among these components are transducers,

amplifiers, filters, multiplexers, sample/holds and analog-to-

digital converters. The system also includes all signal paths

tying these functions together.

Several system architectures will be considered, followed by

a general discussion aimed at the designer who must

choose hardware for a given application.

Topics include:

• Data Acquisition System Architecture

• Signal Conditioning

• Transducers

• Single-Ended vs. Differential Signal Paths

• Low-Level Signals

•Filters

• Programmable Gain Amplifier

• Sampling Rate

• Computer Interface

Data Acquisition System Architecture

At present the most widely used DAS configuration is that

shown in Figure 1. It handles a moderate number of analog

channels, feeding into a common multiplexer, programmable

gain amplifier (if required) track/hold amplifier and A-D

converter.

A more specialized and expensive variation is to place a

Track/Hold in each channel as shown in Figure 2. Switching

all channels to HOLD simultaneously produces a “snapshot”

view which preserves the phase relation of signals in all

channels. This information is important in seismic studies

and vibration analyses.

The DAS system of Figure 3 offers many advantages, but is

not yet practical except for slowly changing channel data.

Low frequency signals allow dedication of a slow but

accurate integrating type A-D converter for each channel.

The channel filters, often included to reduce aliasing errors

and noise, are not necessary, since aliasing is not a problem

with low bandwidth signals. The integrating converter

suppresses wideband noise by averaging it about the

instantaneous signal level. Also, the converter’s integration

period may be chosen to provide almost complete rejection

of a specific interference frequency such as 60 Hz. Digital

outputs from the converters are then digitally multiplexed.

The system shown in Figure 3 has an inherent advantage

over the other two systems, having eliminated both the

track/hold and the analog multiplexer with their many error

contributions. The disadvantage, of course, is cost. Figure 3

would become the system of choice in many more

applications, if a significant reduction should occur in the

price of successive - approximation A-D converters.

A small RAM may be added at the converter’s output in any

of these systems, to buffer the computer and offload its

involvement with individual conversions. Timing and control

may be arranged to scan all channels repeatedly, and

continuously update a RAM location for each channel. The

computer is then free to look at a recent reading for any

channel, at any time.

Further discussion will center on Figure 1, both in the single-

ended version shown, and in the differential version.